Purpose: To establish a relationship between regional changes in lung gas exchange properties and radiation therapy (RT) dose.
Methods: Six patients receiving conventional (2Gy/fx) RT for lung cancer underwent breath-hold hyperpolarized (HP)-¹²⁹Xe gas exchange MRI. Subjects were imaged pre- and 3-months post-RT. Gas exchange MRI utilizes the solubility and chemical shift of ¹²⁹Xe to facilitate 3D imaging of xenon in the lung airspaces (ventilation), alveolar interstitial barrier, and red blood cells (RBC). Voxel-wise ¹²⁹Xe signal ratios of barrier/airspace and RBC/airspace were computed to quantify the interstitial barrier uptake and RBC gas transfer properties of the regional lung tissue, the latter representing an end-to-end measure of function since it is a surrogate for oxygen transfer from alveoli to the bloodstream. Sequential MRI were rigidly registered, and deformed back to the planning CT. We excluded voxels in areas of high dose gradient (≥3Gy/mm) and those with low ventilated xenon signal (signal-to-noise ratio <5), and binned voxels according to the RT dose received (5Gy bin-width). The per-voxel changes in ventilation, barrier uptake, and RBC gas transfer were computed via registered image subtraction. In addition, whole-lung average changes in HP-¹²⁹Xe-measured function were calculated for comparison against changes in the pulmonary function tests (PFTs), forced vital capacity (FVC) and diffusing capacity for carbon monoxide (DLCO).
Results: A least squares fit of data between 5-50Gy showed that ventilation decreased linearly (R²=0.96) by 0.25% Gy⁻¹ between pre- and 3-month post-RT imaging, barrier uptake increased linearly (R²=0.94) by 0.74% Gy⁻¹ (as expected from alveolar interstitial thickening post-RT), and RBC transfer decreased by 0.48% Gy⁻¹ (R²=0.73). Whole-lung increases in barrier signal correlated significantly with decreases in both ΔFVC (R²=0.88, P<0.01) and ΔDLCO (R²=0.71, P<0.04).
Conclusion: Hyperpolarized-¹²⁹Xe MRI can detect RT-induced, dose-dependent changes in regional lung function and may prove useful in diagnosing, grading, and quantifying RILI.
Funding Support, Disclosures, and Conflict of Interest: This work was supported in part by grant R01HL105643 National Institutes of Health (NIH). Bastiaan Driehuys is CTO for, a board member of, and a shareholder in Polarean, Inc.